arxiv: 2604.07090 · v1 · submitted 2026-04-08 · 💻 cs.IR

Recognition: no theorem link

Leveraging Artist Catalogs for Cold-Start Music Recommendation

Yan-Martin Tamm , Gregor Meehan , Vojt\v{e}ch Nekl , Vojt\v{e}ch Van\v{c}ura , Rodrigo Alves , Johan Pauwels , Anna Aljanaki

Authors on Pith no claims yet

Pith reviewed 2026-05-10 17:44 UTC · model grok-4.3

classification 💻 cs.IR

keywords cold-start recommendationmusic recommendationcollaborative filteringartist catalogattention mechanismsemi-cold startACARecitem popularity estimation

0 comments

The pith

New tracks get collaborative filtering embeddings by attending to their artist's existing catalog, more than doubling recall and NDCG.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Music recommendation systems struggle with new tracks that have no user interaction history. This paper reframes the problem as semi-cold-start by exploiting the fact that most new tracks come from artists who already have collaborative data available. Artist-aware methods are shown to more than double recall and NDCG over pure content baselines. The proposed ACARec model uses attention to create collaborative embeddings for new tracks directly from the artist's existing catalog.

Core claim

Since most new tracks come from artists with previous history, cold-start track recommendation can be reframed as semi-cold. Artist-aware methods more than double Recall and NDCG compared to content-only baselines. ACARec generates CF embeddings for new tracks by attending over the artist's existing catalog, offering advantages in predicting preferences for new tracks and estimating cold item popularity.

What carries the argument

ACARec, an attention-based architecture that generates collaborative filtering embeddings for new tracks by attending over the artist's existing catalog.

If this is right

Artist-aware methods more than double Recall and NDCG compared to content-only baselines for cold tracks.
ACARec improves prediction of user preferences for new tracks from known artists.
The approach yields more accurate estimation of cold item popularity.
It supports better new artist discovery by leveraging existing collaborative signals at the artist level.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Platforms could prioritize collecting and maintaining artist-level interaction data to handle new releases more effectively.
The hierarchical attention idea might transfer to other domains with creator histories, such as books by author or videos by channel.
Hybrid models combining artist-catalog attention with audio or metadata features could add robustness when artist data is sparse.

Load-bearing premise

The collaborative signal present at the artist level transfers reliably to new tracks from the same artist, and most new tracks come from artists with prior history.

What would settle it

Measure performance on a test set of tracks from artists with no prior catalog at all; if artist-aware gains vanish and results match content-only baselines, the semi-cold transfer assumption fails.

Figures

Figures reproduced from arXiv: 2604.07090 by Anna Aljanaki, Gregor Meehan, Johan Pauwels, Rodrigo Alves, Vojt\v{e}ch Nekl, Vojt\v{e}ch Van\v{c}ura, Yan-Martin Tamm.

**Figure 2.** Figure 2: Test interactions split on Yambda-50m. cannot make conclusions about general distributions of hot and cold artists; however, as shown in [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Figure 3: Gains from adding ArtistMean into baseline models. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

**Figure 4.** Figure 4: Predictive behavior for ACARec and cold-start baselines across interaction and artist popularity quintiles. A higher [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗

**Figure 5.** Figure 5: Discovery Recall@20 and ACARec’s improvement [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗

read the original abstract

The item cold-start problem poses a fundamental challenge for music recommendation: newly added tracks lack the interaction history that collaborative filtering (CF) requires. Existing approaches often address this problem by learning mappings from content features such as audio, text, and metadata to the CF latent space. However, previous works either omit artist information or treat it as just another input modality, missing the fundamental hierarchy of artists and items. Since most new tracks come from artists with previous history available, we frame cold-start track recommendation as 'semi-cold' by leveraging the rich collaborative signal that exists at the artist level. We show that artist-aware methods can more than double Recall and NDCG compared to content-only baselines, and propose ACARec, an attention-based architecture that generates CF embeddings for new tracks by attending over the artist's existing catalog. We show that our approach has notable advantages in predicting user preferences for new tracks, especially for new artist discovery and more accurate estimation of cold item popularity.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Artist catalog attention doubles cold-start metrics in music recs but transfer reliability needs checking.

read the letter

The key point is that by treating new tracks from known artists as semi-cold starts and using attention to aggregate their catalog's collaborative signals, the authors get more than double the recall and NDCG over content baselines with their ACARec model. This is new in the way it explicitly uses the artist-item hierarchy rather than just adding artist metadata as another feature. Most prior cold-start work in music either ignores artists or flattens them in. The attention architecture lets the model weigh different catalog tracks when creating an embedding for the new one, which fits the data structure. The paper does a good job showing practical gains, particularly for discovering new artists and estimating popularity of cold items. If the datasets are large-scale real logs and the baselines include strong content and CF methods, this could be a useful incremental advance for the subfield. The main soft spot is whether the artist-level signal actually transfers to new tracks. The results assume that interaction patterns across an artist's existing tracks align with what users will do with the new one. Without per-artist performance breakdowns or analysis of catalog embedding variance, it's hard to tell if the big gains come from artists with consistent catalogs or hold more generally. The abstract mentions advantages but the full experiments need to address this to be convincing. This paper is for people building music recommendation systems who face cold-start issues daily. A practitioner or researcher in IR or recsys would find the framing and model details worth considering. It is not reshaping the field but solves a specific problem with a clear method. I would recommend sending it for peer review. The core claim is falsifiable and the architecture is simple enough to implement and test, so referees can check the details and suggest improvements on the validation side.

Referee Report

2 major / 2 minor

Summary. The paper addresses the item cold-start problem in music recommendation by reframing it as a 'semi-cold' task that exploits artist-level collaborative filtering signals. It proposes ACARec, an attention-based model that generates CF embeddings for new tracks by attending over an artist's existing catalog, and reports that artist-aware methods more than double Recall and NDCG relative to content-only baselines, with particular gains for new artist discovery and cold-item popularity estimation.

Significance. If the empirical claims hold under rigorous validation, the work would be significant for cold-start recommendation because it directly leverages the artist-item hierarchy that is ubiquitous in music data but often treated as just another feature. The attention mechanism for proxy embedding generation is a clean architectural idea that could generalize beyond music if intra-artist preference consistency can be demonstrated.

major comments (2)

[Abstract] Abstract: the headline claim that artist-aware methods 'more than double' Recall and NDCG is presented without any experimental details, baseline definitions, dataset descriptions, statistical tests, or ablation results. Because this quantitative improvement is the central empirical support for the semi-cold framing and for ACARec, the absence of these elements makes the primary result unverifiable from the manuscript as written.
[Abstract / Method] The core modeling assumption—that collaborative signals at the artist level transfer reliably to new tracks from the same artist—is load-bearing for both the method and the reported gains, yet no direct evidence (e.g., intra-artist embedding variance, per-artist performance breakdowns, or correlation analysis across catalog tracks) is supplied to test this transfer. If intra-artist preference correlation is low for many artists, the attention-based proxy embeddings would be unreliable and the doubling of metrics could be driven by a small subset of homogeneous catalogs.

minor comments (2)

[Abstract] The abstract states that 'most new tracks come from artists with previous history' but provides no supporting statistic or dataset analysis to quantify this prevalence.
[Method] Notation for the attention mechanism and the mapping from artist catalog to track embedding should be introduced with explicit equations rather than prose description alone.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback. We address each major comment below and describe the revisions we will make to strengthen the manuscript.

read point-by-point responses

Referee: [Abstract] Abstract: the headline claim that artist-aware methods 'more than double' Recall and NDCG is presented without any experimental details, baseline definitions, dataset descriptions, statistical tests, or ablation results. Because this quantitative improvement is the central empirical support for the semi-cold framing and for ACARec, the absence of these elements makes the primary result unverifiable from the manuscript as written.

Authors: We agree that the abstract, as a concise summary, would be more informative if it provided minimal context for the headline claim. The full details—including dataset description, baseline definitions, ablation studies, and statistical significance tests—are presented in Sections 3 and 4 of the manuscript. In the revised version we will expand the abstract to briefly reference the dataset, the content-only baselines, and the fact that reported improvements are statistically significant. This change will improve verifiability without exceeding typical abstract length constraints. revision: yes
Referee: [Abstract / Method] The core modeling assumption—that collaborative signals at the artist level transfer reliably to new tracks from the same artist—is load-bearing for both the method and the reported gains, yet no direct evidence (e.g., intra-artist embedding variance, per-artist performance breakdowns, or correlation analysis across catalog tracks) is supplied to test this transfer. If intra-artist preference correlation is low for many artists, the attention-based proxy embeddings would be unreliable and the doubling of metrics could be driven by a small subset of homogeneous catalogs.

Authors: We acknowledge that direct validation of the intra-artist transfer assumption would strengthen the paper. While the reported gains, particularly the advantages for new artist discovery, provide indirect support for the assumption holding on average, the manuscript does not include explicit analyses such as intra-artist embedding variance or per-artist breakdowns. In the revision we will add these analyses, including intra-artist CF embedding similarity statistics and performance stratified by artist catalog characteristics, to directly test the assumption and address the possibility of subset-driven effects. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation is empirical and self-contained

full rationale

The paper proposes ACARec, an attention-based model that generates CF embeddings for new tracks by attending over an artist's existing catalog, and reports empirical gains in Recall/NDCG over content-only baselines. No equations, derivations, or first-principles results are present that reduce claimed predictions to fitted inputs by construction. The core claim rests on external evaluation against baselines and the observation that most new tracks come from artists with prior history, which is a domain fact rather than a self-referential definition. No self-citations are load-bearing for uniqueness theorems or ansatzes, and the architecture is presented as a novel combination rather than a renaming of known results. The derivation chain is therefore independent of its own outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that artist-level collaborative signals are transferable to new tracks and that artist history is typically available.

axioms (1)

domain assumption Most new tracks come from artists with previous history available
Explicitly stated as the basis for reframing cold-start as semi-cold.

pith-pipeline@v0.9.0 · 5492 in / 1056 out tokens · 47236 ms · 2026-05-10T17:44:16.006203+00:00 · methodology

discussion (0)

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

Sparse Contrastive Learning for Content-Based Cold Item Recommendation
cs.IR 2026-04 unverdicted novelty 7.0

SEMCo uses sparse entmax contrastive learning for purely content-based cold-start item recommendation, outperforming standard methods in ranking accuracy.

Reference graph

Works this paper leans on

82 extracted references · 17 canonical work pages · cited by 1 Pith paper · 3 internal anchors

[1]

Pablo Alonso-Jiménez, Xavier Favory, Hadrien Foroughmand, Grigoris Bourdalas, Xavier Serra, Thomas Lidy, and Dmitry Bogdanov. 2023. Pre-Training Strategies Using Contrastive Learning and Playlist Information for Music Classification and Similarity. http://arxiv.org/abs/2304.12257

work page arXiv 2023
[2]

Haoyue Bai, Min Hou, Le Wu, Yonghui Yang, Kun Zhang, Richang Hong, and Meng Wang. 2023. Gorec: a generative cold-start recommendation framework. InProc. of the 31st ACM international conf. on multimedia. 1004–1012

2023
[3]

Niels Bertram, Jürgen Dunkel, and Ramón Hermoso. 2023. I am all EARS: Using open data and knowledge graph embeddings for music recommendations.Expert Systems with Applications229 (2023), 120347

2023
[4]

Matej Bevec, Marko Tkalčič, and Matevž Pesek. 2024. Hybrid music recommen- dation with graph neural networks.User Modeling and User-Adapted Interaction 34, 5 (2024), 1891–1928

2024
[5]

Rodrigo Borges and Marcelo Queiroz. 2023. Audio-Based Sequential Music Recommendation. In2023 31st European Signal Processing Conference (EUSIPCO). IEEE, 421–425

2023
[6]

Léa Briand, Théo Bontempelli, Walid Bendada, Mathieu Morlon, François Rigaud, Benjamin Chapus, Thomas Bouabça, and Guillaume Salha-Galvan. 2024. Let’s get it started: fostering the discoverability of new releases on Deezer. InEuropean Conference on Information Retrieval. Springer, 286–291

2024
[7]

Yukuo Cen, Jianwei Zhang, Xu Zou, Chang Zhou, Hongxia Yang, and Jie Tang
[8]

InProceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining

Controllable multi-interest framework for recommendation. InProceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining. 2942–2951
[9]

Hao Chen, Zefan Wang, Feiran Huang, Xiao Huang, Yue Xu, Yishi Lin, Peng He, and Zhoujun Li. 2022. Generative adversarial framework for cold-start item recommendation. InProc. of the 45th International ACM SIGIR Conf. on Research and Development in Information Retrieval. 2565–2571

2022
[10]

Ke Chen, Beici Liang, Xiaoshuan Ma, and Minwei Gu. 2021. Learning audio embeddings with user listening data for content-based music recommendation. InICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 3015–3019

2021
[11]

Kyunghyun Cho, Bart Van Merriënboer, Dzmitry Bahdanau, and Yoshua Ben- gio. 2014. On the properties of neural machine translation: Encoder-decoder approaches.arXiv preprint arXiv:1409.1259(2014)

work page Pith review arXiv 2014
[12]

Xiaohui Cui, Xiaolong Qu, Dongmei Li, Yu Yang, Yuxun Li, and Xiaoping Zhang
[13]

doi:10.3390/electronics12122688

MKGCN: Multi-Modal Knowledge Graph Convolutional Network for Music Recommender Systems.Electronics12, 12 (2023). doi:10.3390/electronics12122688

work page doi:10.3390/electronics12122688 2023
[14]

Angelo Cesar Mendes da Silva, Diego Furtado Silva, and Ricardo Marcondes Mar- cacini. 2024. Artist Similarity based on Heterogeneous Graph Neural Networks. IEEE/ACM Transactions on Audio, Speech, and Language Processing(2024)

2024
[15]

Yashar Deldjoo, Dietmar Jannach, Alejandro Bellogin, Alessandro Difonzo, and Dario Zanzonelli. 2024. Fairness in recommender systems: research landscape and future directions.User Modeling and User-Adapted Interaction34, 1 (2024)

2024
[16]

Yashar Deldjoo, Markus Schedl, and Peter Knees. 2024. Content-driven music recommendation: Evolution, state of the art, and challenges.Computer Science Review51 (2024), 100618. doi:10.1016/j.cosrev.2024.100618

work page doi:10.1016/j.cosrev.2024.100618 2024
[17]

Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers). 4171–4186

2019
[18]

Andres Ferraro, Xavier Favory, Konstantinos Drossos, Yuntae Kim, and Dmitry Bogdanov. 2021. Enriched music representations with multiple cross-modal contrastive learning.IEEE Signal Processing Letters28 (2021), 733–737

2021
[19]

Bruce Ferwerda, Eveline Ingesson, Michaela Berndl, and Markus Schedl. 2023. I don’t care how popular you are! investigating popularity bias in music recom- mendations from a user’s perspective. InProceedings of the 2023 conference on human information interaction and retrieval. 357–361

2023
[20]

Christian Ganhör, Marta Moscati, Anna Hausberger, Shah Nawaz, and Markus Schedl. 2024. A Multimodal Single-Branch Embedding Network for Recommen- dation in Cold-Start and Missing Modality Scenarios. InProceedings of the 18th ACM Conference on Recommender Systems. 380–390

2024
[21]

Florian Grötschla, Luca Strässle, Luca A Lanzendörfer, and Roger Wattenhofer
[22]

Towards Leveraging Contrastively Pretrained Neural Audio Embeddings for Recommender Tasks.arXiv preprint arXiv:2409.09026(2024)

work page arXiv 2024
[23]

Jordan, and Nikhil Garg

Wenshuo Guo, Karl Krauth, Michael I. Jordan, and Nikhil Garg. 2021. The Stereotyping Problem in Collaboratively Filtered Recommender Systems.Equity and Access in Algorithms, Mechanisms, and Optimization(2021)

2021
[24]

Ruining He and Julian McAuley. 2016. VBPR: visual bayesian personalized ranking from implicit feedback. InProc. of the AAAI conf. on artificial intelligence

2016
[25]

Feiran Huang, Yuanchen Bei, Zhenghang Yang, Junyi Jiang, Hao Chen, Qijie Shen, Senzhang Wang, Fakhri Karray, and Philip S Yu. 2025. Large Language Model Simulator for Cold-Start Recommendation. InProceedings of the Eighteenth ACM International Conference on Web Search and Data Mining. 261–270

2025
[26]

Feiran Huang, Zefan Wang, Xiao Huang, Yufeng Qian, Zhetao Li, and Hao Chen
[27]

Aligning Distillation For Cold-start Item Recommendation. InProc. of the 46th International ACM SIGIR Conf. on Research and Development in Information Retrieval (SIGIR ’23). Association for Computing Machinery, New York, NY, USA
[28]

Yitong Ji, Aixin Sun, Jie Zhang, and Chenliang Li. 2023. A critical study on data leakage in recommender system offline evaluation.ACM Transactions on Information Systems41, 3 (2023), 1–27

2023
[29]

Jinri Kim, Eungi Kim, Kwangeun Yeo, Yujin Jeon, Chanwoo Kim, Sewon Lee, and Joonseok Lee. 2024. Content-based Graph Reconstruction for Cold-start Item Recommendation. InProc. of the 47th International ACM SIGIR Conf. on Research and Development in Information Retrieval. 1263–1273

2024
[30]

Adam: A Method for Stochastic Optimization

Diederik P. Kingma and Jimmy Ba. 2014. Adam: A Method for Stochastic Opti- mization.CoRRabs/1412.6980 (2014)

work page internal anchor Pith review Pith/arXiv arXiv 2014
[31]

Anastasiia Klimashevskaia, Dietmar Jannach, Mehdi Elahi, and Christoph Trat- tner. 2024. A survey on popularity bias in recommender systems.User Modeling and User-Adapted Interaction34, 5 (2024), 1777–1834

2024
[32]

Filip Korzeniowski, Sergio Oramas, and Fabien Gouyon. 2022. Artist Similarity for Everyone: A Graph Neural Network Approach.Transactions of the International Society for Music Information Retrieval(Oct 2022). doi:10.5334/tismir.143

work page doi:10.5334/tismir.143 2022
[33]

Wo Jae Lee, Rifat Joyee, Emanuele Coviello, and Sudev Mukherjee. 2025. Mul- timodal music tokenization with residual quantization for generative retrieval. (2025)

2025
[34]

Chao Li, Zhiyuan Liu, Mengmeng Wu, Yuchi Xu, Huan Zhao, Pipei Huang, Guoliang Kang, Qiwei Chen, Wei Li, and Dik Lun Lee. 2019. Multi-interest network with dynamic routing for recommendation at Tmall. InProceedings of the 28th ACM international conference on information and knowledge management. 2615–2623

2019
[35]

Dawen Liang, Minshu Zhan, and Daniel PW Ellis. 2015. Content-Aware Collab- orative Music Recommendation Using Pre-trained Neural Networks.. InISMIR. 295–301

2015
[36]

Xinqiao Liu, Zhisheng Yang, and Jinyong Cheng. 2024. Music recommendation algorithms based on knowledge graph and multi-task feature learning.Scientific Reports14, 1 (2024), 2055

2024
[37]

Paul Magron and Cédric Févotte. 2022. Neural content-aware collaborative filtering for cold-start music recommendation.Data Mining and Knowledge Discovery36, 5 (2022), 1971–2005

2022
[38]

Brian McFee, Luke Barrington, and Gert Lanckriet. 2012. Learning content similarity for music recommendation.IEEE transactions on audio, speech, and language processing20, 8 (2012), 2207–2218

2012
[39]

Gregor Meehan and Johan Pauwels. 2025. Artist Considerations in Offline Eval- uation of Music Recommender Systems. MuRS 2025: 3rd Music Recommender Systems Workshop, September 22nd, 2025

2025
[40]

Gregor Meehan and Johan Pauwels. 2025. Evaluating Contrastive Methodologies for Music Representation Learning Using Playlist Data. InICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). 1–5. doi:10.1109/ICASSP49660.2025.10888157

work page doi:10.1109/icassp49660.2025.10888157 2025
[41]

Gregor Meehan and Johan Pauwels. 2025. On Inherited Popularity Bias in Cold- Start Item Recommendation. InProc. of the Nineteenth ACM Conf. on Recommender Systems. 649–654

2025
[42]

M Jeffrey Mei, Florian Henkel, Samuel E Sandberg, Oliver Bembom, and Andreas F Ehmann. 2025. Semantic ids for music recommendation. InProceedings of the Nineteenth ACM Conference on Recommender Systems. 1070–1073

2025
[43]

Zaiqiao Meng, Richard McCreadie, Craig Macdonald, and Iadh Ounis. 2020. Ex- ploring data splitting strategies for the evaluation of recommendation models. In Proceedings of the 14th acm conference on recommender systems. 681–686

2020
[44]

Julien Monteil, Volodymyr Vaskovych, Wentao Lu, Anirban Majumder, and Anton Van Den Hengel. 2024. Marec: Metadata alignment for cold-start recommendation. InProceedings of the 18th ACM Conference on Recommender Systems. 401–410

2024
[45]

Marta Moscati, Emilia Parada-Cabaleiro, Yashar Deldjoo, Eva Zangerle, and Markus Schedl. 2022. Music4All-Onion–A Large-Scale Multi-faceted Content- Centric Music Recommendation Dataset. InProceedings of the 31st ACM Interna- tional Conference on Information & Knowledge Management. 4339–4343

2022
[46]

Sergio Oramas, Andres Ferraro, Alvaro Sarasua, and Fabien Gouyon. 2024. Talking to Your Recs: Multimodal Embeddings For Recommendation and Retrieval. (2024)

2024
[47]

Sergio Oramas, Oriol Nieto, Mohamed Sordo, and Xavier Serra. 2017. A deep multimodal approach for cold-start music recommendation. InProceedings of the 2nd workshop on deep learning for recommender systems. 32–37

2017
[48]

Alexander Ploshkin, Vladislav Tytskiy, Alexey Pismenny, Vladimir Baikalov, Evgeny Taychinov, Artem Permiakov, Daniil Burlakov, and Eugene Krofto. 2025. Yambda-5B—A Large-Scale Multi-Modal Dataset for Ranking and Retrieval. In Proceedings of the Nineteenth ACM Conference on Recommender Systems. 894–901

2025
[49]

Michael Pulis and Josef Bajada. 2021. Siamese Neural Networks for Content-based Cold-Start Music Recommendation.. InProceedings of the 15th ACM conference on recommender systems. 719–723

2021
[50]

Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Tran, Jonah Samost, et al
[51]

UMAP ’26, June 08–11, 2026, Gothenburg, Sweden Tamm and Meehan et al

Recommender systems with generative retrieval.Advances in Neural Information Processing Systems36 (2023), 10299–10315. UMAP ’26, June 08–11, 2026, Gothenburg, Sweden Tamm and Meehan et al

2023
[52]

Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme
[53]

BPR: Bayesian personalized ranking from implicit feedback. InProc. of the Twenty-Fifth Conf. on Uncertainty in Artificial Intelligence (UAI ’09)
[54]

Rebecca Salganik, Xiaohao Liu, Yunshan Ma, Jian Kang, and Tat-Seng Chua. 2024. Larp: Language audio relational pre-training for cold-start playlist continuation. InProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 2524–2535

2024
[55]

Igor André Pegoraro Santana, Fabio Pinhelli, Juliano Donini, Leonardo Catharin, Rafael Biazus Mangolin, Valéria Delisandra Feltrim, Marcos Aurélio Domingues, et al. 2020. Music4all: A new music database and its applications. In2020 In- ternational Conference on Systems, Signals and Image Processing (IWSSIP). IEEE, 399–404

2020
[56]

Andrew I Schein, Alexandrin Popescul, Lyle H Ungar, and David M Pennock
[57]

Methods and metrics for cold-start recommendations. InProc. of the 25th annual international ACM SIGIR conf. on Research and development in information retrieval. 253–260
[58]

Noam Shazeer. 2020. Glu variants improve transformer.arXiv preprint arXiv:2002.05202(2020)

work page internal anchor Pith review Pith/arXiv arXiv 2020
[59]

Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, and Jeff Dean. 2017. Outrageously large neural networks: The sparsely-gated mixture-of-experts layer.arXiv preprint arXiv:1701.06538(2017)

work page internal anchor Pith review Pith/arXiv arXiv 2017
[60]

Anima Singh, Trung Vu, Nikhil Mehta, Raghunandan Keshavan, Maheswaran Sathiamoorthy, Yilin Zheng, Lichan Hong, Lukasz Heldt, Li Wei, Devansh Tandon, et al. 2024. Better generalization with semantic ids: A case study in ranking for recommendations. InProceedings of the 18th ACM Conference on Recommender Systems. 1039–1044

2024
[61]

Janne Spijkervet and John Ashley Burgoyne. 2021. Contrastive learning of musical representations.arXiv preprint arXiv:2103.09410(2021)

work page arXiv 2021
[62]

Changfeng Sun, Han Liu, Meng Liu, Zhaochun Ren, Tian Gan, and Liqiang Nie
[63]

LARA: Attribute-to-feature adversarial learning for new-item recommen- dation. InProc. of the 13th international conf. on web search and data mining
[64]

Yan-Martin Tamm and Anna Aljanaki. 2024. Comparative Analysis of Pretrained Audio Representations in Music Recommender Systems. In18th ACM Confer- ence on Recommender Systems (RecSys ’24). ACM, 934–938. doi:10.1145/3640457. 3688172

work page doi:10.1145/3640457 2024
[65]

Yan-Martin Tamm, Rinchin Damdinov, and Alexey Vasilev. 2021. Quality Metrics in Recommender Systems: Do We Calculate Metrics Consistently?. InFifteenth ACM Conference on Recommender Systems (RecSys ’21). ACM, 708–713. doi:10. 1145/3460231.3478848

work page arXiv 2021
[66]

April Trainor and Douglas Turnbull. 2023. Popularity Degradation Bias in Local Music Recommendation.arXiv preprint arXiv:2309.11671(2023)

work page arXiv 2023
[67]

Robin Ungruh, Karlijn Dinnissen, Anja Volk, Maria Soledad Pera, and Hanna Hauptmann. 2024. Putting Popularity Bias Mitigation to the Test: A User-Centric Evaluation in Music Recommenders. InProceedings of the 18th ACM Conference on Recommender Systems. 169–178

2024
[68]

Aaron Van den Oord, Sander Dieleman, and Benjamin Schrauwen. 2013. Deep content-based music recommendation.Advances in neural information processing systems26 (2013)

2013
[69]

Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need.Advances in neural information processing systems30 (2017)

2017
[70]

Maksims Volkovs, Guangwei Yu, and Tomi Poutanen. 2017. Dropoutnet: Ad- dressing cold start in recommender systems.Advances in neural information processing systems30 (2017)

2017
[71]

Dongjing Wang, Xin Zhang, Yuyu Yin, Dongjin Yu, Guandong Xu, and Shuiguang Deng. 2023. Multi-view enhanced graph attention network for session-based music recommendation.ACM Transactions on Information Systems42, 1 (2023), 1–30

2023
[72]

Jianling Wang, Haokai Lu, and Minmin Chen. 2024. Fresh content recommenda- tion at scale: A multi-funnel solution and the potential of LLMs. InProceedings of the 17th ACM International Conference on Web Search and Data Mining. 1186– 1187

2024
[73]

Wenbo Wang, Bingquan Liu, Lili Shan, Chengjie Sun, Ben Chen, and Jian Guan
[74]

Preference Aware Dual Contrastive Learning for Item Cold-Start Recom- mendation. InProc. of the AAAI Conf. on Artificial Intelligence, Vol. 38. 9125–9132
[75]

Yinwei Wei, Xiang Wang, Qi Li, Liqiang Nie, Yan Li, Xuanping Li, and Tat-Seng Chua. 2021. Contrastive learning for cold-start recommendation. InProceedings of the 29th ACM international conference on multimedia. 5382–5390

2021
[76]

He Weng, Jianjiang Chen, Dongjing Wang, Xin Zhang, and Dongjin Yu. 2022. Graph-Based Attentive Sequential Model With Metadata for Music Recommen- dation.IEEE Access10 (2022), 108226–108240. doi:10.1109/ACCESS.2022.3213812

work page doi:10.1109/access.2022.3213812 2022
[77]

Neil Zeghidour, Alejandro Luebs, Ahmed Omran, Jan Skoglund, and Marco Tagliasacchi. 2021. Soundstream: An end-to-end neural audio codec.IEEE/ACM Transactions on Audio, Speech, and Language Processing30 (2021), 495–507

2021
[78]

Weizhi Zhang, Yuanchen Bei, Liangwei Yang, Henry Peng Zou, Peilin Zhou, Aiwei Liu, Yinghui Li, Hao Chen, Jianling Wang, Yu Wang, et al. 2025. Cold-start rec- ommendation towards the era of large language models (llms): A comprehensive survey and roadmap.arXiv preprint arXiv:2501.01945(2025)

work page arXiv 2025
[79]

Xu Zhao, Yi Ren, Ying Du, Shenzheng Zhang, and Nian Wang. 2022. Improv- ing item cold-start recommendation via model-agnostic conditional variational autoencoder. InProc. of the 45th International ACM SIGIR Conf. on Research and Development in Information Retrieval. 2595–2600

2022
[80]

Zhihui Zhou, Lilin Zhang, and Ning Yang. 2023. Contrastive collaborative filtering for cold-start item recommendation. InProc. of the ACM Web Conf. 2023

2023

Showing first 80 references.